MCP · Model Context Protocol

Connect Claude, ChatGPT & Gemini to your data.

Eformogi exposes an MCP server so any MCP-capable agent can read your vault, create consent grants, and act on your behalf — governed by a per-category policy you control. No SDK. No tokens to manage. No surprises.

Get Your API KeyRead the Setup

Setup in 3 steps

1

Create a personal API key

Open Eformogi → AI Agent Access → Manage API Keys. Name it after your client (e.g. "Claude Desktop"). Copy the key — it starts with efg_ and is shown only once.

Open the app →
2

Set your policy

In the same panel, click Policy on your key. For each of the 18 categories choose Auto (agent reads freely), Ask (queues for your approval), or Never (blocked). Defaults are conservative — SSN is Never, medical is Ask, basic_info is Auto.

3

Wire it into your agent

For Claude Desktop, add this to your config file:

{
  "mcpServers": {
    "eformogi": {
      "url": "https://eformogi.com/api/mcp",
      "headers": {
        "Authorization": "Bearer efg_YOUR_KEY_HERE"
      }
    }
  }
}

Config location: macOS ~/Library/Application Support/Claude/claude_desktop_config.json · Windows %APPDATA%\Claude\claude_desktop_config.json

For ChatGPT, Gemini, or custom agents — point your MCP client at https://eformogi.com/api/mcp with the same Bearer header.

Available tools

Six tools, all enforced against your policy. Call tools/list for full JSON-RPC schemas.

read_profileRead vault data for specified categories. Filtered by your policy.
list_consentsList active consent receipts — who you have shared with.
create_consentCreate a consent grant. "Ask" categories queue for your approval.
check_policyShow what categories the agent can access (auto / ask / never).
log_actionLog an explicit action description into the activity feed.
search_orgsSearch organizations the user can share data with.

Permissions model

Every key has an 18-category policy. Each category can be set to one of three rules:

Auto
Agent can read or act freely. Use for low-sensitivity categories like basic_info.
Ask
Each request queues for your approval inside the app. Approve once, the agent retries.
Never
Hard block. The agent gets a denial and cannot retry without a new key.

Every call is logged in Agent Activity. Revoking a key invalidates it immediately — no propagation delay.

Verify your key (curl)

List the tools your key can see:

curl -X POST https://eformogi.com/api/mcp \
  -H "Authorization: Bearer efg_YOUR_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/list"
  }'

Check what your policy allows:

curl -X POST https://eformogi.com/api/mcp \
  -H "Authorization: Bearer efg_YOUR_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": {
      "name": "check_policy",
      "arguments": {}
    }
  }'

FAQ

What is MCP?
Model Context Protocol — an open JSON-RPC 2.0 standard from Anthropic that lets AI agents (Claude, ChatGPT, Gemini, custom agents) call into external tools. Eformogi exposes an MCP server so any MCP-capable agent can read your vault and request consents, governed by your policy.
Does this give the agent unrestricted access to my data?
No. Each personal API key has an 18-category policy. Categories are auto (agent can read), ask (queues for your approval in-app), or never (blocked). Defaults are conservative — SSN/medical/legal are "never" or "ask" out of the box.
How do I revoke access?
Open Eformogi → AI Agent Access → Manage API Keys → Revoke. The key is invalidated immediately; any in-flight agent call fails on the next request.
Where do I see what the agent did?
Every MCP call is logged in agent_execution_trace and surfaced in Agent Activity inside the app. Approvals you grant or deny are also logged with your decision.
Is the protocol public?
Yes. JSON-RPC 2.0 over HTTPS POST to https://eformogi.com/api/mcp. tools/list returns the full schema. No client SDK required.

Ready to connect your agent?

Create a free account, generate a key, and you'll be calling MCP tools in under two minutes.

Get StartedRead the Agents Page
← Home|Open App|Developer Docs|Agent-Verifiable Records

Claude is a trademark of Anthropic, PBC. ChatGPT is a trademark of OpenAI, Inc. Gemini is a trademark of Google LLC. Model Context Protocol is an open standard published by Anthropic. Eformogi is an independent product and is not affiliated with, endorsed by, or sponsored by Anthropic, OpenAI, or Google.